SberMobile Deploys Fedot.ai AI Supervisor
The system will support the company's outsourced contact center operations.
AI assistants are evolving at remarkable speed. Until recently, they were largely limited to functioning as basic chatbots configured to handle simple, repetitive tasks. Today, digital assistants are increasingly being used as business management tools. They no longer just answer questions – they can also monitor service quality.
Service Quality Gets a Major Upgrade
SberMobile, Russia's virtual mobile operator, recently announced that its outsourced contact center has deployed the Fedot.ai AI supervisor. The system was developed by Tsifrovaya Logistika (Digital Logistics). It monitors how managers interact with customers who purchased SIM cards through online marketplaces.
"The solution selects the most suitable neural network for each business process based on the best balance between cost and performance, with priority given to the business version of the GigaChat model. That allows us to analyze 100% of customer interactions without adding human resources. All customer data is stored in a secure, isolated environment, ensuring confidentiality," SberMobile explained.
The company says the AI supervisor will streamline numerous operational processes. Previously, customer inquiries were assigned to managers manually. That approach consumed considerable time and often resulted in uneven workload distribution across employees.
More Automation
Fedot.ai performs comprehensive automated analysis of every customer interaction while eliminating human bias from performance evaluations. The system automatically reviews all customer conversations, identifies issues and notifies employees through corporate messaging channels. At the same time, the algorithm generates recommendations for addressing detected shortcomings. That enables teams to identify customer-service mistakes immediately and continuously improve service quality.
How Does It Work?
What fundamentally distinguishes Fedot.ai from traditional analytics platforms is its ability to build an analytical infrastructure tailored to each customer's needs.
"The AI agent studies incoming data – phone calls, chats, spreadsheets, CRM exports and information from other systems – then independently proposes the processing logic, defines metrics and builds dashboards. The agent automatically discovers relationships across different data sources: it understands how to match a call recording with CRM data, connect a manager's name in a spreadsheet with their messaging history, and interpret metadata originating from different systems. Fedot.ai is designed to process massive data volumes. When analyzing thousands of calls and conversations, it uncovers patterns that remain invisible even during comprehensive manual review," said Andrey Sharypov, Head of Product for Fedot.ai.
Over time, similar AI supervisors could realistically be adopted by contact centers in other industries, including banks, insurance companies and delivery services – anywhere customer-service automation creates value. Companies have been actively searching for optimal solutions in this area over the past several years.
AI Becomes a Trusted Partner
In 2024, Sber reported that its GigaChat neural network was analyzing conversations with corporate clients and determining, with up to 80% accuracy, whether managers communicated politely and professionally. It was one of the first large-scale examples of a large language model being used to evaluate employee performance.
Equally telling are the 2024 figures released by Rostelecom Contact Center. Over a 12-month period, artificial intelligence was deployed in 46% of its projects, processing 38 billion minutes of speech. Customer response times improved sevenfold.
It is increasingly clear that contact center automation will continue to accelerate. At the same time, there is little reason to expect AI assistants to replace human employees entirely. Instead, they are likely to evolve as support tools. They will handle routine work, generate recommendations and surface insights, while human operators focus on the complex cases that require judgment, empathy and direct human involvement.
